Open Devil-SX opened 1 month ago
Maybe we can check if the --timing argument has been successfully enabled for design at first? the building logs may show the status.
Thanks for you quick reply. I think argument has been successfully enabled because without argument compile process would fail and with argument it can be compiled successfully.
that's an unusual behavior, may I have your minimal code for reproduce?
This is a minimal example for reproducing. delay.zip
If you run the current version or delete the timescale ...
of Delay.v
, you can't see any delay on the waveform.
If you change 1ps/1ps
to 1ns/1ps
, you can't see any change of output signal.
What is the default period of InitClock
?
I also tried add --timescale
argument to verilator, but I didn't find correct form.
The default period for Verilator is 1ps/1ps. Each Step()
will push the clock for 1ps by default (the last one, 1ns/1ps will be 1ps).
So if 1ns/1ps is changed, #10 will delay 10ns, which is 10000 ps and needs 10000 Step()
.
I will try to reproduce in the next few days, thanks for you artifacts.
I'm sorry, but I couldn't reproduce this situation in my environment. It would be appreciated if you could provide more details about your environment.
such as verilator, python, picker, gcc, linux, environment variable version
use Verilator backend,
timing
argumentI use some delay assignments in my RTL code. such as
a <= #(10) b
Didn't see any delay on waveform. Verilator version 5.024 supports timing features.